Interoperability of Medication Classification Systems: Lessons Learned Mapping Established Pharmacologic Classes...

Area of Science:

• Pharmacology
• Medical Informatics
• Computational Linguistics

Background:

• Interoperability challenges persist among diverse medication classification systems.
• Accurate mapping is crucial for medication data standardization and clinical decision support.

Purpose of the Study:

• To evaluate the feasibility and accuracy of mapping Established Pharmacologic Classes (EPCs) to SNOMED CT using automated methods.
• To compare the performance of lexical and instance-based mapping approaches against an expert-reviewed standard.

Main Methods:

• Lexical and instance-based automated methods were employed to map 543 EPCs to SNOMED CT.
• Performance was assessed using precision, recall, and F1 scores against a reference standard.
• Contributions of each automated method were evaluated.

Main Results:

• 284 EPCs mapped directly, 205 were more specific, and 54 remained unmapped.
• Lexical mapping achieved precision=0.416, recall=0.620, F1=0.498.
• Instance-based mapping achieved precision=0.616, recall=0.504, F1=0.554.

Conclusions:

• Both automated methods provide unique strengths and weaknesses for mapping medication classifications.
• Combining automated mapping with expert review enhances identification of matches, gaps, and quality improvements.
• Subject matter expert review remains essential for validating final mappings.